Professional Documents
Culture Documents
Textbook An Introduction To Kac Moody Groups Over Fields Marquis Ebook All Chapter PDF
Textbook An Introduction To Kac Moody Groups Over Fields Marquis Ebook All Chapter PDF
https://textbookfull.com/product/algebra-1-groups-rings-fields-
and-arithmetic-1st-edition-ramji-lal/
https://textbookfull.com/product/quantum-theory-groups-and-
representations-an-introduction-1st-edition-peter-woit-auth/
https://textbookfull.com/product/introduction-to-the-theory-of-
lie-groups-1st-edition-roger-godement/
https://textbookfull.com/product/introduction-to-averaging-
dynamics-over-networks-1st-edition-fabio-fagnani/
From Groups to Categorial Algebra Introduction to
Protomodular and Mal tsev Categories Compact Textbooks
in Mathematics Bourn
https://textbookfull.com/product/from-groups-to-categorial-
algebra-introduction-to-protomodular-and-mal-tsev-categories-
compact-textbooks-in-mathematics-bourn/
https://textbookfull.com/product/an-introduction-to-hinduism-
gavin-flood/
https://textbookfull.com/product/an-introduction-to-integral-
transforms-patra/
https://textbookfull.com/product/an-introduction-to-indian-
philosophy-perrett/
https://textbookfull.com/product/an-introduction-to-feminism-
lorna-finlayson/
Textbooks in Mathematics
Timothée Marquis
Timothée Marquis Timothée Marquis
An Introduction to
An Introduction
The book assumes only a minimal background in linear algebra and basic topology,
and is addressed to anyone interested in learning about Kac–Moody algebras and/or
groups, from graduate (master) students to specialists.
ISBN 978-3-03719-187-3
www.ems-ph.org
Marquis Cover | Font: Frutiger_Helvetica Neue | Farben: Pantone 116, Pantone 287 | RB 30 mm
EMS Textbooks in Mathematics
An Introduction
to Kac–Moody
Groups over Fields
Author:
Timothée Marquis
IRMP
Université Catholique de Louvain
Chemin du Cyclotron 2
1348 Louvain-la-Neuve
Belgium
E-mail: timothee.marquis@uclouvain.be
Key words: Kac–Moody groups, Kac–Moody algebras, infinite-dimensional Lie theory, highest-
weight modules, semisimple algebraic groups, loop groups, affine group schemes, Coxeter groups,
buildings, BN pairs, Tits systems, root group data
ISBN 978-3-03719-187-3
The Swiss National Library lists this publication in The Swiss Book, the Swiss national bibliography,
and the detailed bibliographic data are available on the Internet at http://www.helveticat.ch.
This work is subject to copyright. All rights are reserved, whether the whole or part of the material is
concerned, specifically the rights of translation, reprinting, re-use of illustrations, recitation, broad-
casting, reproduction on microfilms or in other ways, and storage in data banks. For any kind of use
permission of the copyright owner must be obtained.
Contact address:
Email: info@ems-ph.org
Homepage: www.ems-ph.org
Typeset using the author’s TEX files: le-tex publishing services GmbH, Leipzig, Germany
Printing and binding: Beltz Bad Langensalza GmbH, Bad Langensalza, Germany
∞ Printed on acid free paper
987654321
To Oliver
Preface
The interest for Kac–Moody algebras and groups has grown exponentially in the
past decades, both in the mathematical and physics communities. In physics, this
interest has essentially been focused on affine Kac–Moody algebras and groups
(see e.g. [Kac90]), until the recent development of M-theory, which also brought
into the game certain Kac–Moody algebras and groups of indefinite type (see
e.g., [DHN02], [DN05], [FGKP18]). Within mathematics, Kac–Moody groups
have been studied from a wide variety of perspectives, reflecting the variety of
flavours in which they appear: as for the group functor SLn , which associates
to each field K the group SLn .K/ D fA 2 Matn .K/ j det A D 1g, Kac–Moody
groups can be constructed over any field K. In addition, Kac–Moody groups come
in two versions (minimal and maximal). To a given Kac–Moody algebra is thus
in fact attached a family of groups, whose nature can greatly vary. Just to give a
glimpse of this variety, here is a (neither exhaustive nor even representative, and
possibly random) list of recent research directions.
Note first that any Kac–Moody group naturally acts (in a nice way) on some
geometric object, called a building. Buildings have an extensive theory of their
own (see [AB08]) and admit several metric realisations, amongst which CAT(0)-
realisations. In turn, CAT(0)-spaces have been extensively studied (see [BH99]).
This already provides powerful machineries to study Kac–Moody groups, and
connects Kac–Moody theory to many topics of geometric group theory.
Over K D R or K D C, minimal Kac–Moody groups G are connected
Hausdorff topological groups. In [FHHK17], symmetric spaces (in the axiomatic
sense of Loos) associated to G are defined and studied. In [Kit14], cohomo-
logical properties of the unitary form K of G (i.e. the analogue of a maximal
compact subgroup in SLn .C/) are investigated. Maximal Kac–Moody groups
over K D C, on the other hand, have a rich algebraic-geometric structure (see
e.g., [Mat88b], [Kum02], [Pez17]).
Over a non-Archimedean local field K, the authors of [GR14] associate spher-
ical Hecke algebras to Kac–Moody groups of arbitrary type, using a variant of
buildings, called hovels.
When K is a finite field, minimal Kac–Moody groups provide a class of dis-
crete groups that combine various properties in a very singular way. For instance,
they share many properties with arithmetic groups (see [Rém09]), but are typi-
cally simple; they in fact provide the first infinite finitely presented examples of
discrete groups that are both simple and Kazhdan (see [CR09]). They also helped
construct Golod–Shafarevich groups that disproved a conjecture by E. Zelmanov
(see [Ers08]).
viii Preface
Maximal Kac–Moody groups over finite fields, on the other hand, provide an
important family of simple (non-discrete) totally disconnected locally compact
groups (see [Rém12], [CRW17], and also 9.4).
Despite the manifold attractions of general Kac–Moody groups, the vast ma-
jority of the works in Kac–Moody theory still focus on affine Kac–Moody groups.
We strongly believe that this is in part due to the absence of an introductory text-
book on the subject (apart from Kumar’s book [Kum02] which, however, only
covers the case K D C), which can make learning about general Kac–Moody
groups a long and difficult journey. The present book was born out of the desire
to fill this gap in the literature, and to provide an accessible, intuitive, reader-
friendly, self-contained and yet concise introduction to Kac–Moody groups. It
also aims at “cleaning” the foundations and providing a unified treatment of the
theory. The targeted audience includes anyone interested in learning about Kac–
Moody algebras and/or groups (with a minimal background in linear algebra and
basic topology — this book actually grew out of lecture notes for a Master course
on Kac–Moody algebras and groups), as well as more seasoned researchers and
experts in Kac–Moody theory, who may find in this book some clarifications for
the many rough spots of the current literature on Kac–Moody groups. A descrip-
tion of the structure of the book, as well as a guide to the reader, are provided at
the end of the introduction.
To conclude, some acknowledgements are in order. I am very much indebted
to Guy Rousseau, first for his paper [Rou16] which made it possible for me to
write Chapter 8 of this book, and second for his thorough comments on an earlier
version of that chapter. I am also indebted to Pierre-Emmanuel Caprace, for intro-
ducing me to the world of Kac–Moody groups in the first place, and for his pre-
cious comments on an earlier version of Chapter 7. Finally, I extend my warmest
thanks to Ralf Köhl and anonymous reviewers for their precious comments on an
earlier version of the book. Needless to say, all remaining mistakes are entirely
mine.
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
II Kac–Moody algebras . . . . . . . . . . . . . . . . . . . . . . . . 35
3 Basic definitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.1 Preliminaries: presentations and enveloping algebra
of a Lie algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.2 Towards Kac–Moody algebras . . . . . . . . . . . . . . . . . . . 39
3.3 Generalised Cartan matrices . . . . . . . . . . . . . . . . . . . . 40
3.4 Gradations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.5 Realisations of GCM and Kac–Moody algebras . . . . . . . . . . 41
3.6 Simplicity of Kac–Moody algebras . . . . . . . . . . . . . . . . . 49
3.7 The invariant bilinear form . . . . . . . . . . . . . . . . . . . . . 51
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 315
Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 321
The story begins with fundamental works of S. Lie, W. Killing, and E. Cartan back
at the end of the 19th century, which led to the creation of Lie theory. Lie’s orig-
inal motivation was to develop a Galois theory for differential equations; by now,
Lie theory has gone far beyond this objective, and has become a central chapter of
contemporary mathematics. At its heart lies the study of certain groups of symme-
try of algebraic or geometric objects (the Lie groups), of their corresponding sets
of infinitesimal transformations (their Lie algebras), and of the fruitful interplay
between Lie groups and Lie algebras (the Lie correspondence).
To any Lie group G, one can namely associate the vector space g of its tangent
vectors at the identity, and equip g with a Lie bracket g g ! g W .x; y/ 7! Œx; y,
turning g into a Lie algebra. For our purposes, it will be sufficient to think of
G D SLn .C/, in which case g D sln .C/ is the Lie algebra of traceless complex
nn matrices, with the Lie bracket given by ŒA; B WD AB BA for all A; B 2 g.
Conversely, the Lie group G can be reconstructed from its Lie algebra g by
exponentiation: the exponential map
X 1
expW sln .C/ ! SLn .C/ W A 7! eA WD An
n0
nŠ
whose parameters are the entries of the corresponding Cartan matrix A. Now,
this presentation still makes sense if one allows more general integral matrices A,
called generalised Cartan matrices. The corresponding Lie algebras (the Kac–
Moody algebras) were introduced independently in 1967 by V. Kac (whose orig-
inal motivation was to classify certain symmetric spaces, see [Kac67], [Kac68])
and R. Moody ([Moo67], [Moo68]). They share many properties with their (finite-
dimensional) older sisters, but also show some striking differences. These differ-
ences account for a very rich theory of Kac–Moody groups (i.e. of groups asso-
ciated to a Kac–Moody algebra), with the apparition of new phenomena that are
absent from the classical theory.
We give below a brief outline of the story that this book is trying to tell, starting
from finite-dimensional simple Lie algebras, and moving towards the construction
of objects deserving the name of “Kac–Moody groups”.
of g on itself.
The first step in trying to understand the structure of g is to prove the existence
of a Cartan subalgebra h of g, namely, of a nontrivial subalgebra h all whose el-
ements h are ad-diagonalisable (i.e. ad.h/ 2 End.g/ is diagonal in some suitable
basis of g) and that is maximal for this property. Then the elements of h are simul-
taneously ad-diagonalisable: in other words, g admits a root space decomposition
M
gD g˛ ; (1)
˛2h
where
g˛ WD fx 2 g j Œh; x D ˛.h/x 8h 2 hg
is the ˛-eigenspace of ad.h/. The nonzero elements ˛ 2 h such that g˛ ¤ f0g
are called roots, and their set is denoted . One shows that g0 D h, so that (1)
may be rewritten as
M
gDh˚ g˛ : (2)
˛2
1 Finite-dimensional simple Lie algebras 3
Example 1. Let g D sln .C/, and write Eij for the n n matrix with an entry “1”
in position .i; j / and “0” elsewhere. The subalgebra
h WD spanC h˛i_ WD Ei i Ei C1;i C1 j 1 i n 1i
of all diagonal matrices in sln .C/ is a Cartan subalgebra: the ad-diagonalisability
of h follows from the computation
Œ˛i_ ; Ej k D .ıij ıi k ıi C1;j Cıi C1;k /Ej k D ."j "k /.˛i_ /Ej k for all i; j; k;
where "j .Ei i / WD ıij . The corresponding set of roots and root spaces are then
given by
D f˛j k WD "j "k j 1 j ¤ k ng and g˛j k D CEj k ;
L
yielding the root space decomposition sln .C/ D h ˚ j ¤k CEj k .
The second step is to establish some properties of the g˛ ’s. Here are some
important ones:
(1) dim g˛ D 1 for all ˛ 2 .
(2) For any nonzero x˛ 2 g˛ (˛ 2 ), there is some x˛ 2 g˛ such that the
assignment
0 1 0 0 _ 1 0
x˛ 7! ; x˛ 7! ; ˛ WD Œx˛ ; x˛ 2 h 7!
0 0 1 0 0 1
defines an isomorphism g.˛/ WD Cx˛ ˚ C˛ _ ˚ Cx˛ ! sl2 .C/ of Lie
algebras.
(3) ˛.ˇ _ / 2 Z for all ˛; ˇ 2 .
The third step is to study the root system and to show that, together with
the integers ˛.ˇ _ / (˛; ˇ 2 ), it completely determines g. Actually, admits
a root basis … D f˛1 ; : : : ; ˛` g (i.e. every ˛ 2 can be uniquely expressed as a
P
linear combination ˛ D ˙ `iD1 ni ˛i for some ni 2 N), and g is already uniquely
determined by the Cartan matrix
A D .aij /1i;j ` WD .˛j .˛i_ //1i;j `:
More precisely, choosing elements ei D x˛i 2 g˛i and fi D x˛i 2 g˛i as
above, g is generated by the ` copies g.˛i / WD Cfi ˚ C˛i_ ˚ Cei of sl2 .C/
(1 i `), and can even be reconstructed as the complex Lie algebra gA on the
3` generators ei ; fi ; ˛i_ and with the following defining relations (1 i; j `):
Œ˛i_ ; ˛j_ D 0; Œ˛i_ ; ej D aij ej ; Œ˛i_ ; fj D aij fj ; Œfi ; ej D ıij ˛i_ ; (3)
.ad ei /1aij ej D 0; .ad fi /1aij fj D 0 for i ¤ j : (4)
Note that the relations (4), called the Serre relations, make sense, as the aij 2 Z
in fact satisfy aij 0 whenever i ¤ j .
4 Chapter 0. Introduction
2 Kac–Moody algebras
To define infinite-dimensional generalisations of the simple Lie algebras (and,
later on, of the simple Lie groups), we follow the opposite path to the one leading
to the classification of simple Lie algebras (and groups): we start from “gener-
alised” Cartan matrices A, then define a Lie algebra associated to A, and then,
eventually, a group associated to this Lie algebra.
More precisely, the presentation of the Lie algebra gA introduced in the previ-
ous section still makes sense if A D .aij /1i;j ` is a generalised Cartan matrix
(GCM), in the sense that, for each i; j 2 f1; : : : ; `g,
(C1) ai i D 2 (to ensure that ei ; fi ; ˛i_ span a copy of sl2 .C/),
(C2) aij is a nonpositive integer if i ¤ j (to ensure that the Serre relations (4)
make sense),
(C3) aij D 0 implies aj i D 0 (because of the Serre relations .ad ei /1aij ej D 0
and .ad ej /1aj i ei D 0).
The resulting Lie algebra gA is the Kac–Moody algebra associated to A (or rather,
its derived Lie algebra, see Chapter 3 for more details).
Another, maybe more illuminating, way to introduce Kac–Moody algebras, is
to ask the following question: which Lie algebras g can one obtain by keeping the
following fundamental properties of finite-dimensional simple Lie algebras:
(KM1) Generation by ` linearly independent copies g.i / g of sl2 .C/ (i 2 I WD
f1; : : : ; `g). Let us write g.i / D Cfi ˚ C˛i_ ˚ Cei , where 0 1ei ; fi0 and
˛i_ WD Œf
1 0i i; e are respectively identified with the matrices 0 0 , 1 0
0
bij fj for some aij ; bij 2 C. Note that, since the elements of h are simulta-
neously diagonalisable, they commute:
In particular, 0 D Œ˛i_ ; Œfj ; ej D .bij C aij /˛j_ , that is, bij D aij for
all i; j 2 I . Together with (5), this implies that for all i; j 2 I ,
WD f˛ 2 h n f0g j g˛ ¤ 0g
If A is of affine type, then gA is of polynomial growth, in the sense that growthA .n/
grows as a polynomial in n for n ! 1. As a result, Kac–Moody algebras of affine
type are still well understood; in particular, they possess explicit realisations as
1
matrix algebras
2
over the ring CŒt; t of Laurent polynomials. For instance, if
A D 2 2 , then gA is isomorphic to (a one-dimensional central extension of)
2
3 Kac–Moody groups
L
Let gA D h ˚ ˛2 g˛ be a Kac–Moody algebra (the notations being as in the
previous section). In trying to reproduce the classical (i.e. finite-dimensional) Lie
theory in our infinite-dimensional setting, the next step is to ask whether one can
construct a group “attached” (in any reasonable sense) to gA . In other words:
How can one construct a group GA deserving the name of “Kac–
Moody group of type A”?
The answer to this question very much depends on the category of groups one
wants to consider: for instance, to the finite-dimensional simple Lie algebra sln .C/,
one can attach the group SLn .C/, which is not just an abstract group, but also a
topological group, and even a Lie group. With a more algebraic perspective, one
can also attach to sln .C/ the group functor SLn associating to each (commutative,
unital) ring k the group
this group functor even has the structure of an affine group scheme (see Ap-
pendix A for a short introduction to these notions).
As we will see in the third part of this book, analogues for each of the above
mentioned group structures associated to sln .C/ can be obtained in the setting of
8 Chapter 0. Introduction
general Kac–Moody algebras, with the exception of a smooth Lie group structure,
which remains at present elusive. On the other hand, new structures that are spe-
cific to the infinite-dimensional setting also arise (for instance, topological group
structures that are only non-discrete when the group is infinite-dimensional).
Depending on the targeted category of groups, the required amount of effort
to construct a “Kac–Moody group” may greatly vary. For instance, an analogue
of SLn .C/ (or rather, PSLn .C/) as an abstract group for an arbitrary Kac–Moody
algebra gA can be defined straightaway. Indeed, keeping the notations of Exam-
ples 1 and 2, we recall that SLn .C/ is generated by its root groups
exponentiating the root spaces of sln .C/. To give a sense to exponentiation and
generation by subgroups for a general Kac–Moody algebra gA , one naturally con-
siders the “ambient space” Aut.gA /. Since by (KM4) (and, more specifically,
P s
(RR2)), the exponentials exp ad x D s0 .adsŠx/ define elements of Aut.gA / for
x 2 g˛ whenever ˛ 2 is a real root, this suggests to define the group
which one might call an “adjoint complex Kac–Moody group of type A”. For
instance, if gA D sln .C/, then re D and 2 G2
A is the image of SLn .C/ in
Aut.gA /, that is, GA Š PSLn .C/. If A D 2 2 as in Example 3, one can
1
check that GA Š PSL2 .CŒt; t /.
The group GA is an example of a minimal Kac–Moody group, in the sense that
it is constructed by only exponentiating the real root spaces of gA . If one also takes
into account imaginary root spaces, one typically obtains a certain completion of
a minimal Kac–Moody group, called a maximal Kac–Moody group.
2 2 1
Example 4. Let A D 2 2 , so that gA D sl2 .CŒt; t / (up to a central exten-
sion by C). One shows that D Z¤0 ı for some imaginary root ı. Moreover,
im
t n 0 the
imaginary root space gnı of gA is spanned by the diagonal matrix xn WD 0 t n .
s ns
Note that ad xn is not locally nilpotent (for instance, .ad xn /s 00 10 D 00 2 0t ¤
0 for all s 2 N), as indicated by (IR2). In other words, if one wishes to exponen-
tiate xn , one has to allow formal power series in t: the exponential exp xn makes
sense in the maximal Kac–Moody group G b A WD SL2 .C..t/// for each n 2 N,
where C..t// denotes the field of complex formal Laurent series.
While the minimal versions of Kac–Moody groups are easier to define, they
are also usually harder to study. On the other hand, the various Kac–Moody
groups one can construct come in very different flavours. For instance, the min-
imal Kac–Moody group GA defined above can be turned into a connected Haus-
dorff topological group; maximal Kac–Moody groups defined over finite fields,
4 Structure of the book and guide to the reader 9
on the other hand, are naturally (non-discrete if A is not of finite type) totally dis-
connected locally compact groups (think of SLn .CŒt; t 1 / versus SLn .Fq ..t///
for Fq a finite field of order q).
In the third part of this book, we explore in detail the following questions:
What are the possible (currently known) constructions of Kac–Moody
groups, and how much additional structure can they be equipped with?
How do these constructions relate to one another; is there a unique “good”
definition of Kac–Moody group?
Along the way, we will encounter one of the most powerful tools to study Kac–
Moody groups: buildings. These are certain simplicial complexes on which Kac–
Moody groups act nicely (an introduction to buildings and groups acting on them
is given in Appendix B). They yield a geometric interpretation of many important
properties of Kac–Moody groups, thereby adding to this algebraic subject a nice
geometric flavour.
5 Conventions
Throughout the book, we denote by N D f0; 1; : : : g the set of nonnegative integers
and by N D f1; 2; : : : g the set of positive integers. As usual, Z, Q, R and
C respectively denote the sets of integers, rational numbers, real numbers and
complex numbers. We further set Z WD Z n f0g.
By a ring we always mean a commutative, unital, associative Z-algebra. For a
ring k, we write k for the set of its invertible
elements.
Finally, if x 2 C and n 2 N, we let xn denote the binomial coefficient
! !
x x.x 1/ .x n C 1/ x
WD if n > 0 and WD 1:
n nŠ 0
Part I
A few words
on the classical Lie theory
Chapter 1
From Lie groups to Lie algebras
Lie theory was created at the end of the 19th century by S. Lie, W. Killing and
E. Cartan. At its core lies the study of Lie groups G (which were originally viewed
as groups of continuous symmetries of some algebraic or geometric object), of
their corresponding Lie algebra (which were then viewed as the set of infinitesimal
transformations of that object), as well as of the Lie correspondence between a Lie
group and its Lie algebra. [The reader who is unfamiliar with the definitions of
Lie groups and algebras may focus on Example 1.3 below, or just skip Chapter 1,
whose purpose is purely motivational.]
We recall that a Lie group is a topological group G with a compatible smooth
manifold structure, in the sense that the multiplication and inversion maps
W G G ! G W .g; h/ 7! gh and W G ! G W g 7! g 1
are smooth. In particular, for each g 2 G, the left multiplication and conjugation
maps
g W G ! G W h 7! gh and cg W G ! G W h 7! ghg 1
S
are smooth as well. If g 2 G, we write Tg G (resp. T G D g2G Tg G) for the
tangent space of G at g 2 G (resp. for the tangent bundle of G). Thus Tg G is
the vector space of equivalenceˇ classes Œ of smooth paths W Œ1; 1 ! G with
d ˇ
.0/ D g; we also write dt t D0 .t/ for Œ . Given a smooth map f W G ! G, we
further write
ˇ
d ˇˇ
Tf W T G ! T G W x 2 Tg G 7! Tf .x/ D Tg f .x/ WD f . .t// 2 Tf .g/ G
dt ˇt D0
d ˇ
ˇ
for the corresponding tangent map, where x D dt t D0
.t/.
For each x 2 T1 G, there is a unique left invariant vector field x` W G ! T G
with x` .1G / D x, defined by x` .g/ WD T1 g .x/. Since the Lie bracket of two
left invariant vector fields is again a left invariant vector field, we get a continuous
Lie bracket on g WD T1 G, characterised by Œx; y` D Œx` ; y` for x; y 2 g. The
resulting Lie algebra L.G/ WD .g; Œ; / is called the Lie algebra of G. Note
that the conjugation automorphisms define an automorphic action of G on its Lie
algebra, called the adjoint action of G, defined by
Remark 1.1. The assignment G 7! L.G/ defines a functor L (the Lie func-
tor) from the category of Lie groups to the category of Lie algebras (see A.1),
16 Chapter 1. From Lie groups to Lie algebras
that is, there exists a smooth curve x in G passing through 1G and which is
tangent, at each instant t, to the left invariant vector field x` . This defines a smooth
map
expG W g ! G W x 7! expG .x/ WD x .1/;
called the exponential map. Moreover, one can show that expG is a local dif-
feomorphism, in the sense that there is a 0-neighbourhood V in g and a 1G -
neighbourhood U in G such that expG jV W V ! U is a diffeomorphism. This
provides a way to go back from the Lie algebra g to the Lie group G (at least
locally): this interplay between a Lie group and its Lie algebra is what we call
the Lie correspondence. In the finite-dimensional context, the Lie correspon-
dence is well-understood: the functor L defines an equivalence of categories from
the category of simply connected finite-dimensional Lie groups to the category of
finite-dimensional real Lie algebras.
Example 1.3. The group G D GLn .R/ equipped with the smooth manifold struc-
2
ture inherited from Matn .R/ Rn is a Lie group. As G is open in Matn .R/,
its tangent space at A 2 G is TA G D Matn .R/, where a matrix B 2 TA G is
represented by the smooth curve .t/ D A C tB in G (for jtj small enough). In
particular, given B 2 g D gln .R/ WD TI G D Matn .R/, the left invariant vector
field B` is given, for all A 2 G, by
ˇ
d ˇˇ
B` .A/ D T1 A .B/ D A.I C tB/ D AB 2 TA G D Matn .R/:
dt ˇt D0
KUVERNÖÖRIN LUONA
Lisäksi vaikutti Morenoon kaikki, mitä hän tähän asti oli nähnyt
uudessa ympäristössään, unelta, erittäin sekavalta unelta, jota järki
ei ollenkaan ymmärtänyt. Assanoff oli kyllä tavallaan selittänyt, mitä
ympärillä tapahtui; mutta paitsi sitä, että insinööri piti kaikkea tätä
aivan luonnollisena ja syrjäytti helposti juuri enimmin selitystä
kaipaavat asiat, oli hän epävakainenpa kykenemätön mitään
selitystä tai ajatuskulkua loppuun viemään.
"No, hyvä", sanoi soturi ylpeästi, "Murad, Hassan Beyn poika, jos
olet isäsi poika, näytä mihin pystyt!"
"Oletko hullu? Mitä merkitsee se ilve, jota äsken esitit? Kuinka voit
sillä tavoin näytellä itseäsi? Etkö voi ilmaista kotiseudunrakkauttasi
muutoin kuin tuollaisilla vinkuroimisilla?"
Hän poistui heti. Kun Assanoff sai nämä sanat sellaiselta kauniilta
tytöltä, jota tähän asti oli pidetty aivan tunteettomana ja kokonaan
voittamattomana, — jossa ruumiillistui kaupungin kaikkien
tanssijattarien maine, juuri siksi, että hän oli hyvin vähän taipuvainen
kykyään näyttämään, palauttivat ne hänet taas yhdellä iskulla
sivistyksen piiriin, jonka hän oli joku minuutti sitten näyttänyt
kokonaan unohtaneen, ja pistäen kätensä Morenon kainaloon hän
veti tämän muutamia askeleita syrjään ja kuiskasi hänen korvaansa:
"Kenen kanssa?"
"Hyvä on, mutta näytät ilman sitäkin tänä iltana jo kyllin olevan
päästäsi pyörällä!"
TATAARIEN KUNNIA
"Oletteko siellä?"
"Kun sinä otat minut vastaan tällä tavalla, niin en oikein tiedä mitä
täällä teen."
"Et siis koskaan näe siitä yöstä unta? Panet maata ja nukahdat ja
lojut kaiketi kuin laiska lihamöhkäle aamuun asti tai ehkäpä
puolipäivään ajattelematta mitään! Ja lopulta ehkä teetkin siinä
oikein! Sinun koko elämäsihän on vain kuolemaa! Et muistele
mitään, et mitään mennyttä? Enosi, isäni, minun isäni — tiedäthän?
Et, sinä et tiedä mitään! No, minä sanon sinulle: minun isäni siis,
Elam Bey, hirtettiin puuhun karavaanipolun vasemmalle puolelle;
sinun isäsi, minun setäni, naulattiin pistimellä kotioveensa. Sitä et kai
enää muista? Tosin olit vasta kaksitoista-vuotias, mutta minä olin
vain neljän vanha enkä ole siitä mitään unohtanut! Ei, en ole, en,
sanon sinulle, en vähintäkään, en pienintäkään seikkaa. Erään
sotilaan kuljettaessa minua ohitse roikkui enosi puussa niinkuin tuo
takki tuolla seinällä takanasi roikkuu naulassa!"
"Odotin vain sitä aikaa, jolloin olisin kyllin vahva! Kuusi kuukautta
vain lisää — ja minä olisin tappanut hänen molemmat tyttärensä!"
"Muistatko sitä?"